OpenAI Introduces GPT-6.2 Astra and Sol: More Work with Fewer Tokens
· AI · Cem Koyluoglu
OpenAI launched the GPT-6 generation with Astra in September. A few weeks later at DevDay, it upgraded Sol to GPT-6.1. Today GPT-6.2 Astra and GPT-6.2 Sol come with ChatGPT Work and...
TL;DR OpenAI has launched GPT‑6.2 Astra and Sol, boosting performance with fewer output tokens and adding an Ultrafast mode for faster generation. The new models are accessible to Plus, Pro, Enterprise, and Edu subscribers via ChatGPT Work, Codex, and the API, with Astra priced 20 % lower than its predecessor and Sol matching Astra’s performance at a fraction of the cost.
Key Highlights • GPT‑6.2 Astra and Sol deliver higher efficiency, completing demanding tasks with fewer tokens. • Astra excels at complex projects and image‑based template matching; Sol offers comparable coding performance at roughly one‑fifth the cost. • Ultrafast mode, available on Pro plans, can generate tokens up to 4.5× faster and is accessible via the API with updated pricing.
OpenAI introduced the GPT‑6 generation in September, launching the Astra model and later upgrading Sol to GPT‑6.1 at DevDay. Today the company released GPT‑6.2 Astra and GPT‑6.2 Sol, which are now available alongside ChatGPT Work and Codex. The firm claims that both new models can complete demanding tasks with fewer output tokens, allowing users to perform more work before reaching their usage limits.
What’s New in GPT‑6.2 Astra and Sol? Astra continues to be positioned as the model recommended for the most challenging projects. For instance, if a user has a screenshot of a slide and wants the rest of a presentation to match, GPT‑6.2 Astra can use that image as a template, matching layout, colors, and even font size. On the Codex side, Astra is said to excel at making changes to an existing project and testing those changes. “First test users reported cleaner coding style, more focused tests, and fewer revisions needed,” the company noted.
Sol is marketed as a more affordable option for everyday coding and writing tasks, and it comes close to Astra in a coding test. FrontierCode 2.0 evaluates whether changes made in real codebases are merge‑ready for project owners. According to OpenAI, when both models are run at the highest reasoning level, the score difference between Sol and Astra does not exceed half a point, and Sol achieves this result at roughly one‑fifth the cost of Astra.
Who Can Access GPT‑6.2? Both models are now available in ChatGPT Work and Codex for Plus, Pro, Enterprise, and Edu subscribers. The $200 monthly Pro plan grants access to the Ultrafast mode with Sol for the first time, while the $500 monthly Pro plan enables Ultrafast for both models. Enterprise and Edu workspaces can also enable the feature once administrators activate it.
Ultrafast mode can produce tokens at up to 4.5 times the standard speed, even if usage exceeds the allotted quota. Users can switch between models within the speed selector.
Developers can now call both models via the API. Astra’s standard input and output pricing is 20 % lower than that of GPT‑6 Astra, while Sol retains the GPT‑6.1 pricing structure.
Nuvem Yönetici 143500 Puan Yazarın Profili